Inspiration
Applying for jobs can be frustrating. Candidates often read lengthy job descriptions, compare them with their resumes, and struggle to identify exactly what they are missing or how to prepare.
Most basic resume tools focus on keyword matching or generic scores. But a keyword appearing in a resume does not necessarily mean a candidate can demonstrate that skill.
We built RoleReady AI to make this process more transparent, practical, and actionable.
What it does
RoleReady AI is an AI-powered career readiness copilot that evaluates a candidate's resume against a specific job description.
Users can upload a resume in PDF, DOCX, or TXT format and paste a job description. The application then generates an evidence-based assessment containing:
- Readiness indicator: A guidance score based on the evidence identified in the resume, not an ATS score or hiring prediction.
- Requirement matching: Classifies individual requirements as Matched, Partial, Missing, or Unclear.
- Resume evidence: Explains what information in the candidate's resume supports each assessment.
- Priority actions: Highlights the most important areas to improve before applying.
- Resume improvements: Provides specific suggestions without inventing candidate experience.
- Interview preparation: Generates role-specific practice questions.
- Preparation plan: Organizes suggested actions into practical next steps.
Our central design principle is simple: Never invent a candidate's skills, experience, or achievements.
How we built it
We developed RoleReady AI as a Python-based Streamlit application with an integrated AI analysis pipeline.
The main technologies include:
- Python: Core application and analysis logic.
- Streamlit: Interactive user interface.
- OpenRouter: Access to configurable language models for resume and job-description analysis.
- PyMuPDF: PDF text extraction.
- python-docx: DOCX resume processing.
- Requests: API communication.
- Pytest: Automated testing.
- Docker: Containerized application packaging.
- GitHub Actions: Automated testing and container build pipeline.
- GitHub Container Registry (GHCR): Docker image storage.
The application extracts resume content, sends the relevant information alongside the job description to the selected AI model, and processes the response into a structured report.
We also implemented response parsing and validation to handle malformed model output more reliably.
Challenges we faced
One major challenge was ensuring that AI-generated assessments remained grounded in the actual resume rather than simply repeating keywords from the job description.
We addressed this by designing explicit evidence rules and requirement classifications.
Another challenge was handling different resume formats. PDF, DOCX, and TXT files require different extraction approaches, and scanned PDFs can require additional vision-model processing.
We also encountered inconsistent structured responses from language models. To improve reliability, we added JSON parsing, validation, and an automatic repair attempt.
Finally, we worked on making the application reproducible and easier to maintain through automated tests and Docker-based packaging.
What we learned
Building RoleReady AI helped us explore practical applications of LLMs beyond simple conversational interfaces.
We learned how to combine document processing, prompt engineering, structured model outputs, API integration, error handling, testing, and containerization into one application.
Most importantly, we learned that useful AI career tools should explain their conclusions and acknowledge missing evidence instead of presenting an unexplained score as a definitive judgment.
What's next
We plan to expand RoleReady AI with:
- Persistent candidate profiles and analysis history.
- More advanced semantic skill matching.
- Multiple resume versions tailored to different roles.
- Improved scanned-document and image-based resume extraction.
- Interview simulation with feedback.
- Additional model providers and configurable analysis strategies.
- More robust cloud deployment and monitoring.
Our long-term goal is to make RoleReady AI an end-to-end career preparation companion that helps candidates understand where they stand and what they can do next.
Built With
- actions
- ai
- analysis
- artificial
- docker
- fastapi
- generative
- ghcr
- github
- intelligence
- llm
- machine-learning
- natural-language-processing
- openrouter
- pymupdf
- pytest
- python
- resume
- streamlit
